strage system behaviour, and kernel crashes
 

ok, heres the gist of things, I'm trying to install a new kernel for a new install of gentoo

i used rsbac-dev-sources (i use rsbac to lighten security for certain programs, to be able to use them from a regular user ...), whenever i startup that kernel, the kernel crashes, with main stuff (unless you speak hexa-computer) is "bad EIP value" and "kernel panic - not syncing: attempted to kill init"

when i save the config, and use it for a regular gentoo-dev-sources it startup perfectly, and then comes strange behaviour, when i tried to install something, 2 times the kernel crashed (i don't remember the error), then the other time i tried it goes ok to finish a package download, but everything after that , whenever it trys to download more packages, it goes "cant save file", "Input/Output error" ....

my question is: what could possibly cause such bizarre behavior? ... both use 2.6.10 kernels, the live cd i used to install the system from works perfectly, i can chroot and update the environment, the only thing different from most others environment is that i use LC_ALL=en_US.utf8 ... this messes with ncurses, but never gave another problem (and i use that env variable when i chrooted to, with no errors)

the only thing i can think of is IO schedulers, i disabled 2 of them (told them not to compile in the config), that way, only the deadline IO scheduler got compiled, could this cause these problems?, if so, is there a way to use the deadline IO scheduler without making the system random?

ok, now this is just getting ridiculous ..... i can compile new kernel easily with the boot disk, but each new kernel has the same problems (except no input/ouput error ... if it can get that still, it doesn't get far enough for me to see it), whenever i run a program, the system just goes out for lunch, and never returns ... i can switch terminals, i can type in letters, and see them on the screen, but i can not log in (it goes out after i enter in my user name, i cant get it to let me enter in a password), and i cant kill the program thats not working (which is "emerge" ... strange enough, sometimes it works, sometimes it doesn't) .. i type ctr-c .. nothing, ctrl-z .. nothing .....

the only way to get a usable system is to just reboot, then i try removing the /usr/portage/distfiles/* (its the location were all the packages are downloaded) .. the system crashes .. if i remove it from a chrooted environment (chroot off the gentoo install live-cd), the kernel oops's, segfault is displayed, and it kicks me out of the chrooted environment (and that is just deleting some files) .. then if i try to delete them from the cd, the same thing happens, only as I'm not in chroot, i get kicked back to login (which is bad because the damn gentoo live cd's have random root passwords .. which is for "security" (yea right ......) ....

anyways, now my computer is has just gone completely to to the gutter .... if it makes any difference, i also added some new hardware just before this happened, a 80gb hard drive (to replace the old 40gb one that was making sounds .,.. now i have 2 80gb hard drives, from the same manufacturer (and when i tried backing up data from the first, and older 80gb drive, to the new one (i wanted to repartition the first drive), the copying of files always failed, saying the drive was full .. only problem was, according to "df" , the drive was only 6% full (1 partition for the entire drive .. it was only meant to be a temporary place for my files) ... i eventually just did "dd /dev/hda /devhdb" and made a exact copy of the entire drive, and it worked fine ) ... the other piece of hardware i added was a tv-tunner card (not sure if its fully supported tho ... i got it for free from a friend) ... anyways,.... does anyone know of what is causing this behavior, or has any ideas? ... i cant seem to figure anything out, (and i tried testing the drives with "hdparm -t /dev/hda" .. no surprise .. the system went out for lunch)

edit: for some reason, this entire problem seems to always happen around disk I/O time .... and i know the hard drive is good, it never failed before)

Tried a Knoppix-CD yet?



Cheers,
Tink

Try memtest86. It's on the Gentoo CD. Run it for a few passes and it will tell you if you have bad memory or not.

it may be what you said (i am far from expert)
On the one hand you are saying that you want the schedular to be authoritarian and meet deadlines, and on the otherhand you are running an extension that is interupting the system calls and
performing extra tasks that havent been taken into consideration by the original deadline guarantees.

One thing i don't know is if the kernel deadline schedular is using hard real-time task times or something softer ? hard set task times could certainly cause what you describe.

hmm. i think ill take that back about the scheduler ... it seems even the 2.4 kernel on the live-cd is effected (i cant delete files without getting a traceback, and it kicking me) .. the one i have on my computer (2.6, one using cfq, the other deadline or the IO scheduler) will also go berserk at the deletion of those files

i think its probably a reiserfs bug, probably related to me using en_US.utf8 as the locale (i probably did something wrong during the install i guess), ill reformat that drive, and see what happens after a new bootstrap of the system ......